JBD

Iowahawk does a nice job of explaining why political polls in general, and "margins of error" in particular, are scientifically bogus, and practically useless as predicitive indicators of a future vote.

If you are a non-math type, and your eyes glaze over reading his analysis, then you have no business being swayed by any news report quoting a poll, since you haven't seen the raw data, and you couldn't interpret it even if you did.

When a news report says "the so-and-so poll shows it 44 to 43 in favor of Obama, with 13 undecided, and a 4.3 % margin of error" they are simply dressing up the pollsters prejudices in words that sound like science. (Same thing they do in "climate change" reports, but that's another rant.) But they know nobody will listen to a report that says "these guys say that the people they called tonight who would give them an answer prefer Obama over McCain."

Polls are skewed (changed) by the questions asked, and by who does the asking, and by who is asked, and when, and by which medium, and so on. Poll accuracy is based entirely on the honesty of the pollster. Which is why politicians spin public polls to their best advantage, but still pay for their own private polls so that they know what they are really up against.

And that margin of error bit is strictly a function of the number of people who responded to the poll. It is insignificant compared to all other factors, not even worth mentioning, but pollsters mention it anyway, to make it sound scientific.

So remember: the only poll that is accurate is the actual vote on election day.

Posted by: JBD at 08:23 PM | No Comments | Add Comment
Post contains 291 words, total size 2 kb.